Assessing Learners
Assessing two things without confusing them
CLIL assessment is dual: it should capture subject learning and language development, but the two should not contaminate each other unfairly. A learner may understand the science fully yet be unable to express it in the target language, so an assessment that requires extended written prose in the additional language may under-report content knowledge. Practical responses include allowing visual or diagrammatic answers, short structured responses, word banks and glossaries, extra time, and clear, simply worded rubrics.
Assessment purposes must be distinguished: diagnostic (before teaching - establishes prior knowledge and language needs), formative (during learning - feedback that moves learners forward, plus self- and peer assessment), and summative (at the end - reporting achievement). Assessment for learning techniques fit CLIL well: can-do statements, learning intentions shared with learners, traffic-light self-checks, exit tickets, portfolios of work showing progress over time, and rubrics with separate criteria for content and language.
Feedback should be specific and actionable, differentiating comments on subject understanding from comments on language. Correcting every language error in a content assessment discourages risk-taking and distorts the picture of subject learning.
Validity and fairness are key concepts: does the assessment measure the intended construct, and is the language demand itself the barrier? Cultural assumptions in texts, idioms and unfamiliar contexts can also add irrelevant difficulty. Standardised subject tests may be inappropriate for CLIL learners without support.
Traps: assuming a language error means a conceptual error; using only summative written tests; treating peer assessment as unreliable and therefore invalid; confusing BICS with CALP when judging performance; ignoring separate language criteria entirely.
Rule of thumb: ask of any assessment, 'Could a learner prove they understand this content without fluent writing?' If not, add alternative evidence.